For example, in order to support a large number of template engines, Pylons uses Buffet, another library which provides a common interface for these engines. Typically, however, you will only use a single template engine in your project and it only needs a few lines of code to set an engine up.
Although I do most of my web development in Django, when I do need more control or something a bit simpler I go with a Werkzeug + SQLAlchemy + Jinja stack (plus WTForms, similar to Django forms but in a standalone package). The code needed to get up and running isn't that much more than Pylons (and paster generates a whole lot of code anyway) and I can count the dependencies on one hand. That makes deployment a lot faster and easier, and it's just as easy to swap out these dependencies as with Pylons.